Incident Triaging Agent
Agentic AI orchestration that listens to incident tickets reported in BMC RemedyForce and triages the ticket through reading documentation and querying our database. Provides users with what action was taken for the user to take the next steps.
Overview
Multi-agent framework that aims to review newly reported incidents and reduce time for human review.
Problem
When tickets come into our system, it requires development and Product Owner time to diagnose and identify the issue at hand. This leads to context switching and commonly leads to existing work being delayed.
Approach
Created requirement specifications to outline the two use cases to address out the gate. Built a detailed product overview that we inserted into a knowledgebase. After this we designed multiple agents to assist with translating incidents to Jira defects, add context to the question by defining source and database table impacts, query our database, and validate the output prior to handing back to a human for further review.
Results
For the few examples we ran through this agentic framework, resolution cycle time went from 1-2 days to under 5 minutes.